RTG 1666 GlobalFood ⋅ Heinrich Düker Weg 12 ⋅ 37073 Göttingen ⋅ Germany www.uni-goettingen.de/globalfood

ISSN (2192-3248)

                                                                      

 www.uni-goettingen.de/globalfood

RTG 1666 GlobalFood

Transformation of Global Agri-Food Systems: Trends, Driving Forces, and Implications for Developing Countries

Georg-August-University of Göttingen

GlobalFood Discussion Papers

No. 68

Food safety standards in the Guatemalan fresh pea sector:

The role of financial literacy in technology adoption

Anna K. Müller Ludwig Theuvsen

May 2015

Food safety standards in the Guatemalan fresh pea sector: The role of financial literacy in technology

adoption

Anna K. Müller*

Ludwig Theuvsen

Georg-August University Göttingen, Department of Agricultural Economis and Rural

Development, Research Training Group “Transformation of the global agri-food system

– Trends, driving forces and implications for developing countries”

*Corresponding author: anna.mueller@agr.uni-goettingen.de

We gratefully acknowledge the support of the DFG (Deutsche Forschungsgemeinschaft)

and Stiftung Fiat Panis. Special thanks go to IFPRI and Globalgap for providing

secondary data.

Abstract

The transformation of the global agrifood system is characterized by the increasing

importance of food safety and quality standards. This trend is challenging farmers in

countries like Guatemala as they lack necessary skills and assets. This study analyzes

the determinants of Globalgap adoption with a special focus on financial literacy. The

concept has not been considered yet in technology adoption studies. Our results indicate

that financial literacy indeed has a significant impact on the probability to adopt the

standard. Our results hold practical implications for development interventions targeting

technology and standard adoption and smallholder market integration.

Keywords: Technology adoption, food safety standards, financial literacy,

developing countries, market integration

JEL Code Q14, Q12, Q16, O33, C38, C31

2

1. Introduction

In international trade with fresh fruit and vegetables, private certification schemes have

become a predominant instrument for assuring food safety and quality (Unnevehr 2000,

Reardon et al. 2009). This development has fueled a controversial debate in research

and practice about the implications for farmers in developing countries. Compliance

with international (and increasingly also national) quality and food safety standards is

seen as an important asset for participating in agricultural value chains. Standard

adoption is associated with more efficient and sustainable production as well as

economic benefits. The relatively low adoption of private food safety standards and the

weakness of public quality assurance institutions in many developing countries remain

in sharp contrast to this tendency.1 This is a concern because non-adoption could lead to

further marginalization of already small, asset-poor farmers (Maertens and Swinnen

2012). In contrast, significant benefits for small farmers may be expected once they

overcome constraints and comply with international food standards (Anders and

Caswell 2007). Hence, identifying factors that favor or constrain the adoption of food

standards is of empirical and practical relevance.

Existing studies stress the role of endowment factors and access indicators in the

standard adoption process. A number of factors - farm land and non-land assets,

collective capital and access to resources like credit, assistance and information - help

farmers undertake the necessary monetary and non-monetary investments (Reardon et

al. 2009; Asfaw et al. 2009; Handschuch et al. 2013; Hansen and Trifković 2014;

Kersting and Wollni 2012; Subervie and Vagneron 2013).

Standard adoption is an investment decision: farmers have to decide how to allocate

their capital, land and family labor. Process standards lead to changes not only in

agricultural production but also in farm management, and complying with the criteria

requires specific financial and managerial abilities. It might be easier for farmers with

more business-related skills, like financial literacy, to comply with food safety standards

as they know how to use the information and adapt to new requirements. Standard

adoption is also often related to credit access. Having a higher level of financial literacy

1 In Guatemala, GlobalGAP is adopted by less than one percent of fresh vegetable producers.

3

might help farmers to use this access more effectively. However, focusing only on

access to resources and endowment factors might not be sufficient to explain the

adoption decision. To the best of our knowledge, there is no empirical literature that

considers the role of financial literacy in technology adoption or standard adoption in

particular. Thus, this study contributes to the standard adoption literature by

investigating whether financial literacy influences technology adoption by small farmers

in developing countries.

In this paper, we examine the case of GlobalGAP adoption among small pea producers

in the Guatemalan highlands. In this sector, sanitary and phytosanitary violations are

prevalent problems and form one of the primary causes of export restrictions. In the

absence of an effective public quality infrastructure, private investment in food safety

and quality has become vital to securing Guatemala’s role as a leading exporter of fresh

peas. We use cross-section data from 277 pea farmers using a control-group design. The

data was collected in 2012 using a stratified random sampling strategy. Descriptive

results indicate that adopters and non-adopters differ in their level of financial literacy.

Estimates from a bivariate probit model suggest that financial skills positively influence

standard adoption. Our results hold practical implications. Acknowledging the

importance of certain skills in the adoption process enables interventions to be more

effectively tailored to bring farmers to adopt standards.

The remainder of the paper is structured as follows. First, we discuss the literature on

the role of financial literacy in investment and technology adoption behavior. Next, we

describe our research background, data and sampling strategy. After introducing our

methodology in section four, we present the descriptive results in section five. In section

six, we present and discuss the econometric results. The paper finishes with our

conclusions.

2. Financial literacy and the adoption of process innovations

Financial literacy

Stated simply, financial literacy describes “a person’s competency to manage money”

(Remund 2010, p. 279). However, financial literacy embraces a variety of dimensions,

such as financial knowledge, communication, financial management and decision

4

making, and planning (ibid.). Financial literacy is a skill that helps individuals better

evaluate their personal economic situation and more effectively wield financial

information in order to make the best possible decision based on their personal situation

and preferences. The concept stresses the importance of the capability of individuals to

use access to financial resources and financial information for their economic well-

being.

Empirical evidence from developed countries suggests a positive relationship between

level of financial literacy and economic decision making. Higher levels of financial

literacy are associated with investments in pension funds (Lusardi and Mitchell 2005),

stock market participation (van Rooij et al. 2011; Almeberg and Widmark 2011;

Behrmann et al. 2010) and investments in financial products (Brown and Graf 2013). In

spite of this relationship, few studies consider financial literacy in the context of

developing countries. Gaurav et al. (2011) show that financial literacy training for

farmers increases the take-up of index-based weather insurance. Drexler et al. (2014)

find that improved financial and management knowledge has a positive effect on

business outcomes among small businesses in the Dominican Republic. Cole and

Sampson (2011) conclude that there is strong correlation between financial literacy and

financial market participation in India and Indonesia.

Education, cognitive skills and technology adoption

In order to understand how financial literacy could affect standard adoption, we

examine the literature on education, cognitive skills and technology adoption. Financial

literacy is assumed to have an effect on technology adoption similar to that of education

and cognitive skills (Gaurav and Singh 2012). The literature on the effect of education

on the adoption of agricultural innovations in developing countries is very broad, and

most adoption studies use education (in years of schooling) as a control variable (Feder

et al. 1981; Foster and Rosenzweig 2010).

So, what are the paths through which education influences behavior? It is useful to

distinguish between the cognitive and non-cognitive effects of education (Appleton and

Bahiluta 1996). The cognitive effects of education embrace the formation of general

skills, such as literacy and the transmission of specific knowledge. Non-cognitive

effects include preferences and changes in attitude (e.g., being open to innovations and

5

changing preferences). Skills like numeracy or literacy help farmers in their everyday

business (e.g., in using inputs based on the recommendations and computing the

adequate dosage for their plots). It also helps them to make planning decisions relating

the economic future of the farm (how to allocate family labor or whether to use a loan

for investment). Non-cognitive effects influence farmers’ attitudes towards new

technologies, among other things.

Through such cognitive and non-cognitive effects, education influences farmers’

allocative ability. Allocative ability is important for adjusting to change (Feder et al.

1981). There is general agreement in the literature on the important role of human

capital in dealing with the disequilibrium effects that result from the introduction of a

new technology (Feder et al. 1981). Foster and Rosenzweig (1995) conceptualize three

concrete channels through which education influences technology adoption. First, more

educated farmers are wealthier farmers and hence are better endowed to adopt new

technologies (income effect). Second, more educated people have better access to

information as their educational level helps them to better gather, process and use

relevant information (information effect). Third, more educated people are better at and

more open towards learning new things, which is essential in technology adoption

(learning effect).

Education measured as attainment in school gives an incomplete picture of the role of

skills and abilities in technology adoption. In many developing countries, schooling

rates are very low or the quality of education is poor (Jolliffe 1998; Hanushek and

Woessmann 2008). Knowledge and skills are mainly acquired through informal

channels (van Rooij et al. 2011). Farmers may learn from their peers, through extension

services, through learning-by-doing or through their cultural background. Examining

the role of skills in innovation adoption better reflects the complexity of education,

schooling and learning. In considering the role of financial literacy in farmers’

innovation adoption behavior, we contribute to a deeper understanding of the role of

human capital in the innovation adoption process of small farmers.

Financial literacy and standard adoption

Considering the aforementioned literature on education and cognitive skills, we assume

financial literacy to affect standard adoption through several channels. First of all,

6

adopting a standard is an investment decision. Farmers have to decide today on how to

allocate farm resources, capital and labor in order to obtain benefits in the future.

Exporters often bear most part of the certification and investment costs for small

farmers. Of course, farmers invest opportunity costs since they attend training and often

have to cope with a more labor intensive production process. Furthermore, exporters

often intend to reduce their support over time, so farmers need to know whether they

have the necessary skills to comply with the standard without support in the future.

Farmers might also be required to undertake some small on-farm investments

themselves, often with the help of a loan. The inputs provided often come in the form of

a loan. So farmers need to have a solid understanding of credit management. Standard

adoption often makes farming more labor intensive, especially since process standards

require the recording and control of all the production processes. When considering

adoption, farmers need to evaluate the economic and financial consequences of standard

adoption for their farms’ economic and financial situation. Proper financial skills are

therefore important for managing food safety and quality standards at the farm level.

In order to understand how financial literacy influences the adoption process, we rely on

the argumentation laid down in the previous section. Financial literacy can have

cognitive and non-cognitive effects. Farmers with better financial skills might have

more capital and credit to undertake the on-farm investments that are sometimes

necessary. High financial literacy is associated with a greater availability of unspent

income and a higher spending capacity (Klapper et al. 2012).

Farmers with high financial literacy learn faster and can use information (e.g., on

required input use) in a more efficient manner. Low levels of financial literacy may

imply higher costs of information gathering (Almeberg and Widmark 2011). Farmers

with better financial literacy skills might learn faster. Farmers with better financial skills

might also have a more positive attitude towards new investments as they are more

confident about their ability to manage change. Non-investment could be a strategy for

avoiding mistakes caused by missing knowledge and skills (Almeberg and Widmark

2011). The better the level of financial literacy, the better a person may be able to

exploit his or her own resources and the more successful that person will be in adopting

innovations in comparison to persons with a lower level of financial literacy. Klapper et

al. (2012) argue that high financial literacy levels come with a better ability to deal with

7

shocks, such as a high inflation rate, an unforeseen change in interest rates, and the

breakdown of an exporter or even a microfinance institution. In summary, farmers with

a higher level of financial literacy have a better allocative ability and are better equipped

to adjust to the disequilibrium that is caused by the introduction of a new technology or

situation.

3. Research background

3.1GlobalGAPandfoodsafetyinGuatemala

GlobalGAP is the most common private food safety standard for fresh fruit and

vegetable trade that affects developing countries. GlobalGAP is a pre-farm gate

standard that requires the implementation of good agricultural practices as well as

quality and food safety measures. This process standard is non-mandatory and applies

exclusively to business-to-business relationships. It is generic in nature and sets norms

that are slightly above the public regulations of the EU and the US. GlobalGAP is

quasi-mandatory for supplying to several big European retail chains. In order to make

GlobalGAP more accessible to small farmers, there are two certification options:

individual certification and group certification. For group certification, producer groups

run a joint quality management system and can share some investments, like collection

centers and auditing costs. In the recertification process, only a random fraction of the

group is audited, which significantly reduces certification costs. Each producer holds a

contract and is obliged to market certified products exclusively through the group (see

GlobalGAP general regulations 2013).

Guatemala is a country with a very low institutional capacity in food safety and quality.

This challenges public and private compliance efforts and increases the costs for

complying with international norms (Henson 2007). Food safety and quality problems

have been widespread (Norton et al. 2003) and are jeopardizing the international

competitiveness of the country in non-traditional agricultural exports (Julian et al.

2000). Pea exports in particular have experienced high detention rates due to

microbiological contamination and pesticide overuse (Henson 2007). These detentions

have considerable economic effects, as the export-oriented sector is dominated by

capital-poor smallholders.

8

For several years, the non-traditional export sector has been using GlobalGAP

increasingly as an instrument to reach conformance with international norms, and today

it is the most important food quality standard for Guatemala. In August 2012 there were

1,233 certified farmers in Guatemala (GlobalGAP 2012). Over 800 of the certificates

are held by pea producers. GlobalGAP-certified production is still marginal: Less than 1

percent of fresh fruit and vegetable producers in the country are certified by

GlobalGAP.

3.2Data

Between August and October 2012, we surveyed a sample of 277 fresh pea farmers in

the departments of Chimaltenango and Sacatepéquez in the Guatemalan highlands.2 We

collected information on the socio-demographic and socio-economic situation of the

farm-households as well as on agricultural production and marketing, certification and

financial literacy. The recall period was from August 2011 to July 2012. The financial

literacy section is based on widely used survey questions (OECD INFE 2011; Atkinson

and Messy 2012). Six multiple choice questions cover general knowledge of numeracy

(percentage calculation and division) and more specific financial knowledge (inflation,

interest and compound interest calculation). We presented the questions as a small quiz

rather than a test to the farmers to make them feel more comfortable. If a farmer was not

able to answer the two general numeracy questions we did not ask them the detailed

financial literacy questions. The test questions were then coded as “does not know”.3

We use a stratified random sample. The treatment group consists of 152 farmers

certified under option 2 (group certified farmers). The first control group consists of 65

non-certified farmers who are members of a farmer group. The selection of the farmer

groups was a non-random process since we had information on farmer groups from

collaborating exporters and one nongovernmental organization. Within the farmer group

we selected certified and non-certified interviewees randomly from the member list.

GlobalGAP certification within the farmer group is still an individual decision. None of

the groups we dealt with for the study had reached full certification of all members.

2 These two departments account for around 90% of the national pea production. They are both relatively close to the capital, Guatemala City, which favors the production of export crops due to better access to modern infrastructure and lower transportation and transaction costs. 3 See Appendix A1 for a detailed presentation of the question used in the financial literacy test.

9

Since some of the information on certification turned out to be outdated, we also have

ex-certified farmers in our sample. Since these farmers decided to adopt in the past, they

are included as adopters in our model. The second control group consists of 60 non-

certified and non-organized farmers. This group sells to intermediaries or on the spot

market, where there is no standardized quality selection of the product. We included this

group to be able to control for group level effects. The second control group was

selected by random walk method.

We use information on transportation costs and distance to the next marketing center;

this data was provided by the International Food Policy Research Institute (IFRPI).

4. Methods

4.1Empiricalmodeloftheadoptionprocess

We model the adoption decision based on a utility maximization framework. We

assume that a utility maximizing farmer opts for GlobalGAP adoption if the expected

utility of adoption is higher than the expected utility of non-adoption. A farmer’s utility

is influenced by socioeconomic and contextual variables that also influence the decision

to adopt GlobalGAP.

The utility function for GlobalGAP adoption takes the following form:

1 ,

where describes the utility of farmer i. is a vector of contextual and socioeconomic

variables assumed to influence utility. We cannot directly observe a farmer’s utility of

adoption as it is a latent variable. What we can actually observe is the farmer’s choice

between adoption and non-adoption. Based on the utility framework, we assume that a

farmer adopts if the utility of GlobalGAP adoption is greater than zero, and does

not adopt if it is not:

2

1 0

0 0.

Assuming a linear relationship, the adoption of GlobalGAP can therefore be described

as

3 ,

10

where is a binary choice variable taking the value 1 if the farmer has adopted

GlobalGAP and 0 if not, is a vector of observed farm and non-farm characteristics

that are assumed to influence the decision, FL is our variable of interest - financial

literacy - and is the unobserved error term we are trying to minimize.

Small farmers have access to GlobalGAP adoption through group certification. Only

those farmers who decide to be members of a group actually have the option of standard

adoption. To acquire consistent estimates, we need to consider the two related

decisions: first, the farmer decides whether to join a farmer group. If the farmer opts for

membership, he or she can decide whether or not to adopt the standard. See figure 1 for

an illustration of the decision process. We assume that non-group members do not face

the decision of GlobalGAP adoption as individual certification entails very high costs.4

It might be the case that the same unobservable factors drive both group membership

and GlobalGAP adoption. We could think of motivation to succeed in economic terms

as a driving factor for group membership and GolbalGAP adoption alike. This could

cause a selection bias. Without correcting for this bias, the model would give

inconsistent estimates.

To control for possible selection bias, we opt for a bivariate probit model, which allows

for correlation in the error term (Greene 2003). If no correlation is found between the

two error terms, then no selection bias exists and two independent probit models can be

4 Among pea producers, the only option 1 certificates are held by exporter-owned farms.

Figure 1 Conceptual framework of GlobalGAP adoption

11

used to interpret the results. If significant correlation exists between the two error terms,

then the bivariate model corrects for the selection bias in the estimates.

The following bivariate model is adapted from Greene (2003) and Kersting and Wollni

(2012):

(4) Selection equation: ∗

, 1 ∗ 0, 0 .

(5) Outcome equation:

∗ , 1 ∗ 0, 0 .

| | 0,

| | 1,

| .

∗ represents the unobserved, latent variables. ∗ is the utility of being in a farmer

group and ∗ is the utility of being certified with GlobalGAP. are parameter vectors,

are vectors of exogenous explanatory variables and are the error terms with zero

mean, unit variance and correlation . The model is tested under the null hypothesis

=0, meaning no correlation between the error terms and no selection on unobservable

factors. If is found to differ significantly from 0, then we have selection bias in our

model.

For our decision model, the following holds:

1 if the farmer i is member of a farmer group, 0 otherwise

1if the farmer i has adopted GlobalGAP, 0 otherwise

We can only observe 1 if 1. Only if a farmer is member of a farmer group

can he or she actually face the adoption decision.

We oversampled GlobalGAP-certified farmers. In order to control for biases caused by

the sampling design, we used probability weights. We used the inverse of the

probability of being included in the sample due to the sampling design. We estimated

the population size of our sampled group based on data provided by GlobalGAP and the

2004 Guatemalan agricultural census (Instituto Nacional de Estadísticas 2004). We use

robust standard errors.

12

4.2Principalcomponentanalysis

Principal component analysis (PCA) is a multivariate statistical method. It is used to

reduce a number of variables that describe the same latent phenomenon into smaller

dimensions. From an initial set of n correlated variables, PCA creates uncorrelated

components. These components account for most of the variance in the data. Each

extracted component is a linearly weighted combination of the initial set of variables.

For a set of variables to the principal components are

(6) ⋯

…

(7) ⋯ ,

where is the weight for the mth component and the nth variable (Vyas and

Kumaranayake 2006). The weights of the principal components are the eigenvectors of

the correlation matrix. The eigenvalue of the eigenvector is the amount of explained

variance (Vyas and Kumaranayake 2006; van Rooij et al. 2011). The first component

accounts for the largest amount of the underlying information of the variables used

(Kolenikov and Angeles 2004). It represents the linear index of all the variables used in

the PCA. The other components are not correlated with the first component. They

explain additional, but smaller, variation in the data. PCA assigns weights to the

variables according to how much each contributes to the variation in the data

(Langyintuo and Mungoma 2008). We used unrotated PCA to construct a financial

literacy index and a farm asset index. Using an index has proved to be useful by other

studies in financial literacy research (van Rooij et al. 2011; Behrmann et al. 2010) and

poverty research (Vyas and Kumaranayake 2006).

For financial literacy, the first extracted component accounts for almost 70% of the

variation (table A-1 in the appendix). The factor loadings for the first component all

have the same sign and are almost equal in magnitude (table A-2, appendix). The

Kaiser-Meyer-Olkin (KMO) criterion of sampling adequacy tests whether the data used

is suitable for PCA (see table A-3 in the appendix). The overall KMO score is higher

than 0.8, which is considered very good. Bartlett’s test of sphericity tests whether the

correlations between the variables used are significant. The test indicates that we can

13

reject the null hypothesis of zero correlations between the variables. We used the first

component to construct the financial literacy index.

The same procedure was applied to the 13 variables associated with farm assets. The

KMO results suggest that we can perform factor analysis, albeit the value of 0.56 is

lower than in the financial literacy index. Bartlett’s test indicates that the data has

enough correlation in order to perform PCA (see tables A-3 to A-5 in the appendix).

The farm asset index is a proxy for the asset endowment of the farm household (as we

do not have the necessary information in our dataset it is not a proxy for wealth).

5. Descriptive results

5.1Samplecharacteristics

In tables 1 and 2, we present the descriptive statistics. We compare the means of

certified and non-certified farmers for several variables of interest and use a t-test to

check whether there are statistically significant differences in mean between the two

groups.

The farmers are mainly indigenous: Only around 6% in either group state that their

mother tongue is Spanish. The main language in the export business and in the (public

or private) extension infrastructure is Spanish. Not speaking proper Spanish might

increase information asymmetries, thus disadvantaging indigenous farmers in the

adoption process.

Almost two-thirds of the farm household members are to some degree involved in

farming activities, which means that we are dealing with family farms. The average off-

farm income per capita in a year is relatively low and does not translate to the minimum

wage per month.5 The average land size and average hectares owned puts both groups

into the category of subsistence farmers with less than seven hectares of land (Instituto

Nacional de Estadísticas 2005).

5 The minimum wage per month in Guatemala in 2012 for non-farm activities was 2,074 Guatemalan quetzals. (See http://www.leylaboral.com/guatemala/hotlinks/salariominimo.htm, accessed 11.11.2014.)

14

Table 1 Sample characteristics

Ever certified Never certified Differences

Farm household and farm characteristics

Age (head) 45.118 43.408 -1.71

Years of education (head) 4.691 4.592 -0.1

Mother tongue (0 = Spanish) 0.059 0.064 0

Total household members 6.217 5.88 -0.34

Members working on farm 3.77 3.656 -0.11

Members working off-farm 1.382 1.384 0

Total off-farm income 10,654.974 9,510.408 -1,144.57

Off-farm income per capita (quetzals) 1,867.106 2,092.827 225.7207

Total farm size in ha 1.644 1.172 -0.47**

Land owned in ha 1.446 0.957 -0.49**

Land owned before 2009 in ha 1.005 0.556 -0.45**

Share of peas in % of productive land 37.589 37.207 -0.38

Land title (0 = no title) 0.783 0.688 -0.09*

Irrigation (0 = no irrigation) 0.224 0.168 -0.06

Irrigation in pea production (0 = no irrigation) 0.204 0.144 -0.06

Farm asset index 0.293 -0.364 -0.66***

N 152 125

Differences in mean significant at * p < 0.10, ** p < 0.05, *** p < 0.01

Certified and non-certified farmers differ significantly in total farm size and in the

amount of land they actually own. On average, certified farmers possess more land and

work on larger farms. A significantly higher share of certified farmers possess an

official land title. A formal land title is important for accessing the formal financial

market and gives security over land holdings. Unsecured property rights might hinder

investments. The two groups do not differ significantly in their technological level

proxied by irrigation. However, only around 20% of the farmers use irrigation, which is

not a very high share.

Certified farmers score significantly better on the asset index. The farm asset index

incorporates various durable farm assets as proxies for the economic situation of the

farm. The higher the score in the asset index, the better endowed the farm. Certified and

non-certified farmers differ significantly in group membership. This is not surprising as

we were targeting group-certified farmers. Non-certified farmers also include

independent farmers. Just comparing group members, certified farmers have been group

members for significantly longer time. This might hint at the role of positive trustful

15

relationships in the certification process. Significantly more certified farmers were

working with an exporter before 2009. We took 2009 as a threshold as GlobalGAP

certification became more widespread afterwards. Certified farmers scored significantly

higher in the financial literacy index, our variable of interest.

Table 2 Sample characteristics continued

Ever certified Never certified Differences

Organization

Farmer group member ( 0 = no member) 0.98 0.52 -0.46***

Time of membership 6.538 4.189 -2.35**

Marketing

Experience with buyer in years 5.183 6.161 0.98

Exporter before 2009 (0= no exporter) 0.428 0.152 -0.28***

Business skills

Financial literacy index 0.391 -0.476 -0.87***

Experience in pea production in years 11.187 12.051 0.86

Access

Distance to the next marketing center in meters

6,616.317 6,374.303 -242.0138

Transportation costs ($/kg) 0.0043417 0.0051357 .000794*

Altitude 2,216.782 2,212.607 -4.18

Savings (0 = no savings) 0.164 0.192 0.03

Remittances (0 = no remittances) 0.059 0.08 0.02

Conditional cash transfer (0 = no CCT) 0.191 0.216 0.03

Access to formal credit (0= no Access) .355 .328 -.027

N 152 125

Differences in mean significant at *p < 0.10, ** p < 0.05, *** p < 0.01

The two groups do not differ significantly in distance to the next marketing center,

whereas interestingly they differ significantly in transportation costs. Non-certified

farmers have significantly higher transportation costs. This could indicate that

certification is somehow related to lower transaction costs. Savings, remittances and

conditional cash transfer are proxies for farm households’ economic situation and

access to financial resources. There are no significant differences in mean between the

groups for these variables. The saving rate seems quite low among the respondents (16–

19%). Only 5% of the respondents receive remittances. This is surprising as the area is

known for having a high migration rate to the United States. But it could be that mainly

16

male family members migrate, and female members are probably not so involved in pea

production and certification activities.

The conditional cash transfer program is designed for needy families. They receive a

small subsidy when they comply with certain criteria, like sending kids to school and

attending regular medical checkups.6 Around 20% of the sample receives this subsidy.

But it seems that not only necessity influences whether a family receives the subsidy;

another factor is whether the public sector is present in the area. Thus, receiving the

subsidy is an incomplete proxy for poverty. We do not see any systematic difference in

access to formal credit between the two groups. Around one third of the sample has

access to loans from formal sources such as banks or microfinance institutions.

Certified and non-certified farmers show no systematic difference in mean in farmer or

farm household characteristics. But when it comes to variables related to farm

characteristics, marketing activity, financial literacy and access, we see systematic

differences between the two groups.

6 http://www.mides.gob.gt/programas-sociales/mi-bono-seguro, accessed 20.06.2014.

Figure 2 Distribution of correct answers in the sample N=277

17

5.2Financialliteracy

Around one-third of the respondents did not answer any of the questions in the financial

literacy test correctly (see figure 2). Among those who managed to answer at least one

question, most of the respondents scored three or four correct answers out of six (18%

respectively). The median score is three correct answers out of six.

In order to better understand the possible influence of financial literacy on standard

adoption, we explore the characteristics of the financially literate farmers in our sample.

We stratify our sample into farmers with high and low financial literacy according to

their scores in the index.7 We use a t-test to compare the differences in mean between

the two groups. The statistically significant differences are presented in table 3.

We see that farmers with a better score in the financial literacy index are on average

younger and have attended more years of school. Maybe the younger farmers did not

only attend more years of school, but also benefitted from higher quality of schooling,

which will have improved their skills in areas important to financial literacy (numeracy,

literacy etc.).

Table 3 Characteristics of farmers with high and low financial literacy

High FL Low FL Differences Age (head) 42.900 45.626 2.73* Education 5.492 3.898 -1.59*** Total off-farm income 11,846.385 8,628.075 -3218.31* assetX 0.182 -0.218 -0.40** Formal credit access 0.300 0.381 0.08 Member farmer group 0.823 0.728 -0.10* Years of membership 7.018 4.645 -2.37*** GlobalGAP 0.646 0.463 -0.18*** Observations 130 147 Differences in mean significant at * p < 0.10, ** p < 0.05, *** p < 0.01

It is also assumed that younger farmers are more open to innovations and changing

practices. It might also be the case that younger farmers have had more experience with

loans or other financial products, which in turn would influence their financial literacy

level. Thus, it might have been easier for younger farmers to acquire financial literacy

skills.

7 The cutoff point is the median: Scores below the median indicate low financial literacy; scores above the median indicate high financial literacy.

18

Farmers in the upper quintile of financial literacy have on average higher off-farm

earnings and score on average better in the asset index. This reflects in part the higher

educational level of highly skilled farmers: Education and skills are seen as strong

determinants of earnings and wealth. In turn, financial literacy might also be influenced

by income and wealth status. Interestingly, the two groups do not differ in any farm

characteristics or access indicators, like access to formal credit. (We do not present the

non-significant differences in table 3 due to space restrictions.) Highly financially

literate farmers have a higher membership rate in farmer groups and length of

membership tends to be greater. Group membership is associated with better access to

information and extension services. There might also be better opportunities to learn

from the experience of others. The GlobalGAP certification rate is also higher among

highly skilled farmers.

The situation of financial literacy in our sample reflects the findings of other studies. In

a review of studies dealing with financial literacy, (Lusardi and Mitchell 2014) find that

low levels of financial literacy are associated with poorer and less educated households.

Older people and women are less literate, self-employed individuals perform better than

employed individuals and individuals living in rural areas tend to have lower financial

literacy than those living in urban areas.

6. Estimation results

6.1DeterminantsofGlobalGAPadoption

The selection equation of the bivariate probit model estimates the probability that a

farmer will join a farmer group. The outcome equation estimates the probability of

GlobalGAP adoption (see table 4). For a detailed explanation of the variables used in

the adoption model see table A-6 in the appendix.

Group membership is positively influenced by the age of the farmer, farm assets and

experience working with an exporter (Exporter before 2009). Per capita off-farm

income, remittances, cell phone use, experience in pea production and transportation

costs negatively influence farm group membership.

We cannot reject the null hypothesis of zero correlation between the error terms of the

two equations since differs significantly from 0. We performed a likelihood ratio test,

19

which confirmed that the bivariate model performs better than two independent models.

Hence we rely on the bivariate model to interpret our results.

Table 4 Bivariate probit model with marginal effects

Group membership

GlobalGAP adoption

Coeff. (s.e) Coeff. (s.e.) Marginal effectsa

(s.e.)

Age 0.0196** (0.00972) 0.0126 (0.00851) 0.0019 (0.00130) Gender -1.039** (0.451) -0.745** (0.380) -0.111* (0.0599) Education 0.0282 (0.0469) -0.00717 (0.0371) -0.000572 (0.00527) MembersOnFarm 0.0143 (0.0408) 0.0146 (0.0366) 0.0021 (0.00514) Off_income -0.000103** (4.15e-05) -3.08e-

05 (2.19e-05) -5.34e-06 (3.58e-06)

Ha owned before 2009

0.0877 (0.0738) 0.0562 (0.0625) 0.0085 (0.00889)

Land title 0.166 (0.226) 0.150 (0.191) 0.0217 (0.0263) Irrigation -0.257 (0.246) -0.0688 (0.218) -0.0123 (0.0301) Remittances -0.862** (0.345) -0.798** (0.346) -0.115** (0.0531) Conditional cash transfer

-0.328 (0.201) -0.255 (0.173) -0.0376 (0.0235)

Cell -0.490* (0.256) -0.317 (0.212) -0.0477 (0.0335) BuyerFFV -0.0111 (0.280) -0.161 (0.243) -0.0212 (0.0350) TarmacRoad -0.241 (0.206) -0.0537 (0.171) -0.0101 (0.0254) FarmX 0.294*** (.880) (0.0945) (.7146) 0.034*** (0.0127) Livestock_NR -0.0366 (0.117) 0.113 (0.106) 0.0142 (0.0153) Mother tongue 0.126 (0.458) 0.00356 (0.419) 0.00209 (0.0598) Exporter before 2009

0.610*** (0.213) 0.624*** (0.167) 0.0893*** (0.0262)

Formal credit access

-0.275 (0.185) 0.190 (0.155) 0.0212 (0.0246)

Experience pea production

-0.0363*** (0.0124) -0.0186 (0.0114) -0.00290* (0.00170)

Specialization 0.00345 (0.00650) 0.00149 (0.00502) 0.000239 (0.000728)T_costs -74.03** (30.45) -46.36** (23.30) -7.007* (3.834) FLX2 0.0591 (0.0477) 0.108*** (0.0399) 0.0149** (0.00667) Constant 2.961*** (0.742) 1.918*** (0.638) rho 1.483*** (0.210) Observations 277 152 277 Wald test of rho=0: chi2(1) = 78.8103 Prob > chi2 = 0.0000 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 a: continuous variables at the mean value, binary variables at modal value

20

Unlike in other adoption studies, we do not find a significant effect from such

socioeconomic variables as age, education, off-farm income, member working on farm

and land size (Subervie and Vagneron 2013; Kersting and Wollni 2012). We find that

male-headed households are less likely to adopt the standard than female-headed

households. This result is contrary to the findings of similar studies, such as

Handschuch et al. (2013).

Receiving remittances from a family member abroad significantly decreases the

likelihood of adoption. This effect might result from the absence of family members of

productive age. The remaining family members might not be productive enough to

engage in certification-based pea production. Another reason could be that farm families

do not see a necessity to upgrade agricultural production but may instead invest in non-

farm activities.

The score in the asset index has a positive effect on GlobalGAP adoption, indicating

that the better equipped a farmer is with farm assets, the more likely he or she is to

adopt the standard ceteris paribus. This hints at a wealth effect also found by other

studies (Kersting and Wollni 2012; Asfaw et al 2009). The farmers who are able to

invest in assets may also be more able to undertake the necessary investments that are

not covered by the exporters. If a farmer was already working with an exporter in 2009,

this also increases the likelihood of adoption. Farmers with more experience in export

markets are used to producing in line with certain quality standards and use this

experience in the adoption process. This result is backed up by a study by FAO (2014):

They identify pre-existing relationships with the export market as one important

determinant of standard adoption. Experienced farmers are more likely to have detailed

information about market requirements and future developments, which might also

point to the role of trust and long-standing relationships in the certification process.

A farmer’s score on the financial literacy index has a significant positive effect on

GlobalGAP adoption. This finding confirms our initial assumption that financial literacy

plays a significant role in the adoption decision. (For a deeper interpretation of the role

of financial literacy, see the next section.) Experience in pea production influences

GlobalGAP adoption significantly. Surprisingly, the effect is negative: Farmers with

more years of experience in pea production are less likely to adopt GlobalGAP. More

experienced farmers might be more conservative and less flexible in applying new

21

methods or reluctant to accept external advice. Transportation costs to the next market

town also have a negative effect on our outcome of interest. The further away and the

more remote a farm is, the less willing the farmer seems to invest in standard

compliance. This result partly confirms results from similar studies, like that of Kersting

and Wollni (2012). We refine the estimation of the distance effect using estimated

transportation costs that take into consideration infrastructure and natural conditions.

High transaction costs outweigh the benefits of certification for more remote farmers.

Export companies may be less present in more remote areas as they also suffer from

higher transaction costs in reaching those areas.

To interpret the magnitude of the estimated coefficients, we calculated the marginal

effects of the probit model. For the bivariate probit model, the marginal effects are

reported as joint probabilities for a success on both stages of the model (Group

membership = 1 and GlobalGAP adoption = 1). For continuous variables the marginal

effects are calculated at the means, and for dummy variables at the modal value of the

variable. In our model female farmers are 11% more likely to be GlobalGAP adopters

compared to male farmers. Receiving remittances decreases the adoption likelihood by

about 11%. A one unit change in the asset score increases the likelihood of GlobalGAP

adoption by 3.4%. If a farmer was working with an exporter in 2009, the adoption

likelihood increases by 9%. Experience in pea production has a negative influence on

adoption. The marginal effect is small: An extra year of experience in pea production

decreases the likelihood by 0.3%. The reported marginal effect of transportation costs is

relatively large. For a one unit increase in the financial literacy index, the likelihood of

adoption increases by 1.5%.

6.2FinancialliteracyandtheadoptionofGlobalGAP

We identified a significant positive effect of financial literacy on GlobalGAP adoption

in the case of Guatemalan fresh pea farmers. To determine the magnitude of the effect,

we interpret the marginal effect of financial literacy. A one-unit increase in the financial

literacy index results in a 1.5% higher probability of adopting GlobalGAP ceteris

paribus. For example, the index ranges from -3.6 to 2.6. If a farmer shifts from the

lowest quartile of financial literacy to the highest (an increase of 3 units in the index),

his or her probability of standard adoption increases by 4.5%. A change from no

22

financial literacy to the maximum level of financial literacy increases the adoption

likelihood by 9%.

While financial literacy has a positive effect on GlobalGAP adoption, we do not find a

significant effect of educational level on standard adoption. This result is interesting: It

seems that the financial literacy test captures different skills than we do by including

years of schooling.8 Our descriptive results show that higher scores in financial literacy

come with on average more years of schooling. We can assume that the skills necessary

for standard adoption do not depend on the years a farmer has attended school.

Schooling quality or informal learning might be important sources of the financial skills

necessary for innovation adoption. Van Rooij et al. (2011) argue that level of schooling

is an incomplete proxy for financial or economic skills. Lusardi and Mitchell (2014) see

financial literacy as a result of human capital investments rather than the simple result

of more years of formal schooling. Studies often do not find a significant effect of years

of schooling on technology adoption since schooling quality is low in developing

countries (Jolliffe 1998; Hanushek and Woessmann 2008).

Our regression result suggests that asset endowment is important for the adoption of

GlobalGAP. We show in the descriptive results that farmers with higher financial

literacy skills tend to score better in the asset index. Although the direction of causality

is not clear (on the one hand, financial literacy might help build up assets and improve a

farmer’s economic situation or, on the other hand, having certain asset level may require

improvement in financial skills), we see in the descriptive statistics that highly skilled

farmers benefit from a better asset endowment, which increases their likelihood of

adoption (income effect). Being a member of a farmer group is a prerequisite for

certification. Membership comes with advantages for farmers in the form of improved

access to extension services, information, inputs, loans etc. Higher financial literacy

may help farmers to better use the information and advice necessary for the adoption

process. Low cognitive skills are associated with higher information costs. Christelis et

al. (2010), for example, find that the association between cognitive skills and stock

market participation is driven by information constraints. Non-cognitive effects of

financial literacy might also influence adoption behavior. Financial literacy might also

8 We also ran the model without the financial literacy index. The result was the same: Education is not significant.

23

influence a farmer’s attitude towards certification schemes, for example, by giving more

importance to planning in business and financial aspects. The findings of Burks et al.

(2009) support our assumptions: They show that cognitive skills influence individuals’

preferences.

Financial literacy could also influence standard adoption through the farmers’ risk

attitudes. Financial literacy might help them build resilience and become less vulnerable

towards external shocks, like fluctuating input and output prices, inflation or interest

rate changes. This ability might lower their risk aversion towards the adoption of new

technologies where the future economic outcome is not clear at the moment of adoption.

We do not include any measure of risk aversion in our model, but other studies confirm

that low cognitive skills are associated with impatience and higher risk aversion

(Dohmen et al. 2007; Burks et al. 2009).

Overall, financial literacy may improve farmers’ ability to cope with the disequilibrium

effect caused by new technologies. They adjust better to change and are therefore more

likely to adopt innovations like the GlobalGAP standard. Financial literacy is not

exogenous in our model. Our results might be biased due to some unobserved

characteristics that influence financial literacy and GlobalGAP adoption alike, such as

intelligence, ambition and diligence. We do not control for this due to the lack of an

adequate instrument. Lusardi and Mitchell (2014) summarize the results of research

papers that use an instrumental variable approach to control for the endogeneity of

financial literacy. In the studies they reviewed, the effect of financial literacy on

economic behavior persists even when implementing an instrumental variable approach.

But there may still be unobservable variables, such as motivation or intelligence, which

bias the results. To overcome this bias, panel data or experimental approaches are

needed to isolate the real effect of financial literacy on economic behavior.

7. Conclusions

The increasingly integrated global food system poses new challenges for smallholders.

Whether small farmers benefit from the changes might depend heavily on their skills

and capacity to adapt to change. It is especially important to comply with food safety

and quality standards in order to participate in the high agricultural value chains. The

objective of this paper was to assess the role of financial literacy in standard adoption.

24

Financial literacy has only recently gained attention in agricultural economic research

and has not been studied yet in relation to process innovation adoption.

This study focuses on fresh pea production in the Guatemalan highlands. This

smallholder-dominated sector has suffered a great deal from sanitary and phytosanitary

violations and pesticide overuse. Nevertheless, compliance with food safety standards

such as GlobalGAP is very low. In our study we show that, in addition to capital

endowment and access factors, financial literacy is a significant factor in the standard

adoption process. Farmers with a higher score on the financial literacy index are more

likely to adopt GlobalGAP than those with lower scores on the test. The results confirm

the assumption that not only access and endowment factors, but also skills like financial

literacy play an important role in technology adoption. Whereas cognitive skills in the

form of financial literacy matter in GlobalGAP adoption, formal school education is not

significant in our setting.

Our results have important practical implications for the public and private actors.

Integrating small farmers from developing and transition economies into the modern

agrifood system is a concern for the public sector, development organizations and

private companies, such as exporters. Huge efforts in the form of extension services,

development projects and public subsidy programs are designed in order to help

farmers. The adoption of new technologies is an integral part of rural development

policies. The public sector may take a leading role in providing the infrastructure,

functioning institutions and securing access, but this is not enough. We showed that

farmers’ ability to use resources and access are important in the adoption process.

Hence, farmers’ capacity building should be an integral part of rural development

policies in Guatemala. Formal schooling may not equip farmers with the skills

necessary to cope with new technologies. Informal learning, learning-by-doing and

learning from others seem to be important in skill development. Education policy

should foster business-related learning through formal education but also informal

learning opportunities like group-based learning through farmer field schools or through

the use of information technologies in extension services. It might also be helpful for

farmers interested in food standard adoption to learn from farmers who are already

certified. Platforms for these services could be capacity-building activities or farmer

field days. In Guatemalan agriculture, private actors, such as exporters, are taking the

25

lead when it comes to the adoption of food safety and quality standards. The extension

services and training they provide is typically centered on agronomic topics. As farmers

are increasingly integrated in complex global value chains, they are transforming from

being subsistence farmers to being entrepreneurial farmers. Our results show that

exporters should also include more business-related capacity building into their

extension repertoire if they seek sustainable conformance with international food

standards.

We acknowledge that we have to interpret our results with care. We did not control for

endogeneity in the form of an omitted variable problem as we lack a valid instrument.

Factors like ambition, intelligence or openness might influence financial literacy and

standard adoption alike. We are aware that our results might have limited external

validity. Nevertheless our work is a first explorative step towards a better understanding

of the role of cognitive skills like financial literacy in agricultural innovation adoption.

Small farmers in developing countries are faced with an ever more complex decision

environment. Being equipped with the necessary skills to make proper decisions is vital.

Better knowledge of financial matters helps farmers to improve their decision-making

ability, their capacity to foresee and adapt to market trends and their resilience and

entrepreneurial independence. We took an initial exploratory step towards a better

understanding of the role of financial literacy in standard adoption. Further research

should deepen the understanding of how financial literacy affects technology adoption.

This could be done by considering different dimensions of financial literacy in the

analysis or by stratifying the sample based on literacy groups. The effect of financial

literacy may also depend on interactions with other variables. To improve the validity of

the results, endogeneity problems should be addressed by such means as randomized

control trials and other experimental approaches. Looking deeper into the sources of

financial literacy - whether developed through formal education or through informal

learning and experience - could help improve the design of training programs.

26

8. References

Almeberg, J.; Widmark, O. (2011): Numeracy, financial literacy and participation in

asset markets. Social Science Research Network. Online

http://ssrn.com/abstract=1756674, accessed 26.06.2014.

Anders, S.; Caswell, J. (2007): Standards-as-Barriers versus Standards-as-Catalysts:

Assessing the Impact of HACCP Implementation on U.S. Seafood Imports. Department

of Resource Economics, University of Massachusetts Amherst Working Paper No. 2007-

7.

Appleton, S.; Bahiluta, A. (1996): Education and Agricultural Productivity: Evidence

from Uganda. In: Journal of International Development 8 (3), 415-444.

Asfaw, S.; Mithöfer, D.; Waibel, H. (2009): Investment in compliance with GlobalGAP

standards: does it pay off for small scale producers in Kenya? In: Quarterly Journal of

International Agriculture 48 (4), 337–362.

Atkinson, A.; Messy, F.-A. (2012): Measuring Financial Literacy: Results of the OECD

/ International Network on Financial Education (INEF) Pilot Study. OECD Working

Papers on Finance, Insurance and Private Pensions, No 15, OECD Publishing. Online:

http://dx.doi.org/10.1787/5k9csfs90fr4-en, accessed 5.12.2014.

Behrmann, J.R; Mitchell, O.; Soo, C.; Bravo, D. (2010): Financial Literacy, Schooling

and Wealth Accumulation. National Bureau of Economic Research Working Paper

Series No. 16452. Online: http://www.nber.org/papers/w16452, accessed 26.06.2014.

Brown, M.; Graf, R. (2013): Financial literacy, household investment and household

debt: Evidence from Switzerland. Swiss Institute of Banking and Finance Working

Papers on Finance No. 13, Online:

https://www.alexandria.unisg.ch/publications/219130, accessed 8.12.2014.

Burks, S.V; Carpenter, J.P; Rustichini, A. (2009): Cognitive skills affect economic

preferences, strategic behavior and job attachment. In: PNAS 106 (19), 7745–7750.

Online: http://www.pnas.org/content/106/19/7745.full.pdf+html, accessed 8.12.2014.

Christelis, D.; Jappelli, T.; Padula, M. (2010): Cognitive abilities and portfolio choice.

In: European Economic Review 54 (1), 18–38.

27

Cole, S.; Sampson, T. Zia B. (2011): Prices or Knowledge? What Drives Demand for

Financial Services in Emerging Markets? In: The Journal of Finance LXVI (4).

Dohmen, T.; Falk, A.; Huffman, D.; Sunde, U. (2007): Are risk aversion and impatience

related to cognitive ability? Institute for the Study of Labor Discussion Papers No.

2735. Online: http://ftp.iza.org/dp2735.pdf, accessed 8.12.2014.

Drexler, A.; Fischer, G.; Schoar, A. (2014): Keeping it simple: Financial Literacy and

Rules of Thumb. In: American Economic Journal: Applied Economics 6 (2), 1–31.

Feder, G.; Just, R.; Silberman, D. (1981): Adoption of Agricultural Innovations in

Developing Countries: A Survey. World Bank Staff Working Paper No. 444.

FAO (2014): Impact of international voluntary standards on smallholder market

participation in developing countries: A review of the literature. Food and Agriculture

Organization of the United Nations: Rome.

Foster, A.; Rosenzweig, M. (2010): Microeconomics of Technology Adoption. In:

Annual Review of Economics 2, 395–424.

Gaurav, S.; Singh, A. (2012): An Inquiry into the Financial Literacy and Cognitive

Ability of Farmers: Evidence from Rural India. In: Oxford Development Studies 40 (3),

358–380.

Gaurav, S.; Cole, S.; Tobacman, J. (2011): Marketing Complex Financial Products

in Emerging Markets: Evidence from Rainfall Insurance in India. In: Journal of

Marketing Research XLVIII, 150–162.

Greene, W. (2003): Econometric Analysis. Prentice Hall: New Jersey.

Handschuch, C.; Wollni, M.; Villalobos, P. (2013): Adoption of food safety and quality

standards among Chilean raspberry producers: Do smallholders benefit? In: Food

Policy 40, 64–73.

Hanushek, E. A.; Woessmann, L. (2008): The Role of Cognitive Skills in Economic

Development. In: Journal of Economic Literature 46 (3), 607–668.

Hansen, H.; Trifković, N. (2014): Food Standards are Good – For Middle-Class

Farmers. In: World Development 56, 226–242.

28

Henson, S.; Blandon J. (2007): The Impact of Food Safety Standards on Export-

Oriented Supply Chains: Case of the Horticultural Sector in Guatemala. Economic

Commission for Latin America and the Caribbean: Santiago de Chile. Online:

http://www.cepal.org/mexico/

capacidadescomerciales/TallerCRActH/Documentos/DOCTO_HENSON_FINAL.pdf,

accessed 12.11.2014.

Instituto Nacional de Estadísticas (2004): IV Censo Nacional Agropecuario. INE:

Guatemala.

Instituto Nacional de Estadísticas (2005): Encuesta Nacional Agropecuaria. INE:

Guatemala.

Jolliffe, D. (1998): Skills, Schooling, and Household Income in Ghana. In: The World

Bank Economic Review 12 (1), 81–104.

Julian, J.; Sullivan, G.; Sánchez, G. (2000): Future Market Development Issues

Impacting Central America's Nontraditional Agricultural Export Sector: Guatemala

Case Study. In: American Journal of Agricultural Economics 82 (5), 1177–1183.

Kersting, S.; Wollni, M. (2012): New institutional arrangements and standard adoption:

Evidence from small-scale fruit and vegetable farmers in Thailand. In: Food Policy 37

(4), 452–462.

Klapper, L. F.; Lusardi, A.; Panos, G.-A. (2012): Financial literacy and the financial

crisis. National Bureau of Economic Research Working Paper Series No. 17930.

Online: http://www.nber.org/papers/w17930 accessed 8.12.2014.

Kolenikov, S.; Angeles, G. (2004): The Use of Discrete Data in PCA: Theory,

Simulations and Applications to Socioeconomic Indices. Online:

http://www.cpc.unc.edu/measure/ publications/wp-04-85, accessed 18.08.2014.

Langyintuo, A. S.; Mungoma, C. (2008): The effect of household wealth on the

adoption of improved maize varieties in Zambia. In: Food Policy 33 (6), 550–559.

Lusardi, A.; Mitchell, O. S. (2014): The Economic Importance of Financial Literacy:

Theory and Evidence. In: Journal of Economic Literature 52 (1), 5–44.

Maertens, M.; Swinnen, J. (2009): Trade, Standards and Poverty: Evidence from

Senegal. In: World Development (37) 1, 161–178.

29

Norton, G. W.; Sanchez, G. E.; Clarke-Harris, D.; Traore, H. K. (2003): Case Study:

Reducing Pesticide Residues on Horticultural Crops. In: Food Safety in Food Security

and Trade (IFPRI), Focus 10, Brief 10 of 17.

OECD INFE (2011): Measuring Financial Literacy: Questionnaire and Guidance Notes

for Conducting an Internationally Comparable Survey of Financial Literacy.

Organisation for Economic Co-operation and Development: Paris.

Reardon, T.; Barrett, C. B.; Berdegué, J. A.; Swinnen, J. F.M (2009): Agrifood Industry

Transformation and Small Farmers in Developing Countries. In: World Development 37

(11), 1717–1727.

Remund, D. L. (2010): Financial Literacy Explicated: The Case for a Clearer Definition

in an Increasingly Complex Economy. In: The Journal of Consumer Affairs 44 (2), 276-

295.

Subervie, J.; Vagneron, I. (2013): A Drop of Water in the Indian Ocean? The Impact of

GlobalGAP Certification on Lychee Farmers in Madagascar. In: World Development

50, 57–73.

Unnevehr, L. (2000): Food safety issues and fresh food product from LDCs. In:

Agricultural Economics 23, 231–240.

van Rooij, M.; Lusardi, A.; Alessie, R. (2011): Financial literacy and stock market

participation. In: Journal of Financial Economics 101 (2), 449–472.

Vyas, S.; Kumaranayake, L. (2006): Constructing socio-economic status indices: how to

use principal components analysis. In: Health Policy and Planning 21 (6), 459–468.

30

9. Appendix

A1 Numeracy and financial literacy test

If there is a possibility of 10% of getting ill, how many persons out of 100 would get ill?

Five persons have bought the winning number in a lottery. The prize is 2,000 quetzals.

How much will each winner receive?

Imagine you had 1,000 quetzals in a savings account. The annual interest rate is 2% (20

quetzals in the first year). After five years, how much will you have in the saving

account if you do not touch the money?

More than 1020 quetzals

Exactly 1020 quetzals

Less than 1020 quetzals

Imagine that your income will double next year. The prices of all the products that you

consume will also double. With your income, how much will you be able to buy next

year?

More than this year

The same as this year

Less that this year

The bank has leant you 3,000 quetzals; the interest rate is 1% every month. If you pay

30 quetzals every month, when will you have paid back the loan?

In less than five years

In less than ten years

Never

Imagine you get a loan of 1,000 quetzals from the bank. Which option is better for you?

To pay 5% interest every month

To pay 24% interest a year

31

A2 Principal Component analysis for financial literacy test

Table A-1 Principal components for financial literacy

Principal components/correlation Number of obs. = 277 Rotation: (unrotated = principal) Rho = 1.0000 Component Eigenvalue Difference Proportion Cumulative Comp1 4.10547 3.30873 0.6842 0.6842 Comp2 0.796745 0.373399 0.1328 0.8170 Comp3 0.423346 0.15087 0.0706 0.8876 Comp4 0.272477 0.0172805 0.0454 0.9330 Comp5 0.255196 0.108431 0.0425 0.9755 Comp6 0.146765 0. 0.0245 1.0000

Table A-2 Factor loadings for component 1

Principal components (eigenvectors) Variable Comp1 Probability skills 0.3610 Division skills 0.3033 Interest 0.4553 Inflation 0.4363 Credit repayment 0.4187 Interest2 0.4524

Table A-3 Kaiser-Meyer-Olkin measure of sampling adequacy

Variable KMO Probability skills 0.8972 Division skills 0.8668 Interest 0.8617 Inflation 0.9122 Credit repayment 0.9283 Interest2 0.8698 Overall 0.8888

Table A-4 Bartlett’s test of sphericity

Chi-square 1163.503Degrees of freedom 15 p-value 0.000

32

A3 Principal component analysis for asset index

Table A-5 Principal components for the asset index

Table A-6 Factor loadings for component 1

Table A-7 Bartlett’s test and KMO

Principal components/correlation Number of obs. = 277 Rotation: (unrotated = principal) Rho = 1.0000 Component Eigenvalue Difference Proportion Cumulative Comp1 1.84616 0.429529 0.1420 0.1420 Comp2 1.41664 0.139036 0.1090 0.2510 Comp3 1.2776 0.0526699 0.0983 0.3493 Comp4 1.22493 0.16598 0.0942 0.4435 Comp5 1.05895 0.0197427 0.0815 0.5249 Comp6 1.03921 0.140667 0.0799 0.6049 Comp7 0.89854 0.0263536 0.0691 0.6740 Comp8 0.872187 0.0525365 0.0671 0.7411 Comp9 0.81965 0.105628 0.0671 0.7411 Comp10 0.714022 0.015416 0.0549 0.8591 Comp11 0.698606 0.0892119 0.0537 0.9128 Comp12 0.609394 0.0852783 0.0469 0.9597 Comp13 0.524116 0.0403 1.0000

Principal component 1 (eigenvectors) Variable Comp1 Car 0.1752 Pickup 0.3797 Motorbike 0.1257 Bike 0.0525 Truck 0.3073 Knapsack sprayer 0.2996 Knapsack manual 0.1785 Irrigation 0.3449 Reservoir 0.3278 Storage silo 0.1088 TV 0.3500 Radio 0.2796 Mobile 0.3895

Bartlett’s test of sphericity Chi-square 1163.503Degrees of freedom 15 p-value 0.000 KMO 0.560

33

A4 Variables used in the adoption model Table A-6 Variables, specification and expected effects of the variables used in the adoption model

Variable Specification Expected effect

Age Age of household head in years + Gender Dummy, 1 if male 0 if female Education Education of household head in years of formal schooling + Members Number of household members working on farm + Off-income Household off-farm income per capita in quetzals + Ha owned in 2009 Total ha with formal property title in 2009 + Land title Dummy, 1 if farmer has any formal land title, 0 otherwise + Irrigation Dummy, 1 if farmer is using irrigation on at least one plot, 0

otherwise +

Remesas Dummy, 1 if household is receiving remittances, 0 otherwise + BonoSeguro Dummy, 1 if household is part of conditional cash transfer

program

Cell Dummy, 1 if farmer is using cell phone, 0 otherwise BuyerFFV Dummy, 1 if there is a buyer for fresh fruit and vegetables in the

village, 0 otherwise +

TarmacRoad Dummy, 1 if the village is connected via tarmac road, 0 otherwise

+

Asset Index of farm assets + LivestockNR Number if Livestock owned Mother tongue Dummy, 1 if mother tongue of the farmer is Spanish, 0

otherwise +

Exporter before 2009

Dummy, 1 if farmer has worked with an exporter before 2009 +

FLX Financial literacy index + Credit_formal Dummy, 1 if farm-household has access to formal credit, 0

otherwise +

Experience pea Experience in pea production in years + Specialization Share of land allocated to pea production in 2011/12 in % + T_costs Transportation costs to the next market measured in dollars per

kg -